Project Summary Antibiotic resistance has become a pressing public health concern due to the rise of pathogenic bacterial strains with mutations that reduce or eliminate the effectiveness of drugs to treat infections. Beta-lactamases produced by some bacteria provide resistance by degrading beta-lactams, one of most widely used class of antibiotics. Originally restricted to penicillins, mutant beta-lactamases that confer resistance to antibiotics including monobactams and most cephalosporins (known as extended spectrum beta lactamases, or ESBLs) are widespread. Clinical isolates are currently characterized for ESBL resistance using inhibition zone tests against a panel of lactam antibiotics, but the results produced by these tests are difficult to standardize and do not translate consistently into clinical practice. In addition, these tests typically produce no information about the genetic basis for the observed resistance, nor the relatedness to other potentially characterized strains. Here, we propose the development of a sequence-based analysis platform and knowledgebase for analyzing molecular signatures of extended-spectrum beta-lactamase resistance that can initially market to health institutions and companies monitoring the spread of EBSL resistance. The platform will consist of an analysis kit that extracts positively-selected variants, beta-lactamase sequences, and other genomic information relevant to the ESBL phenotype from whole genome sequences of clinical samples. A total of 662 samples will be analyzed; of which, 350 ESBL-resistant samples provided by the Mercy Center at UC Merced will be newly sequenced and the rest will be obtained from a published study from the University of Washington. A cloud-based, searchable database with interactive visualization will be served as the repository of the ESBL-resistant features identified in the clinical samples. By leveraging the metadata exchange standards being developed for broad sharing of human genomic data, the rapidly expanding Global Alliance for Genomics and Health application program interface, our work will represent the initial extension of this API for sharing microbial-centric data. The ultimate goal of this project is to create an accurate, predictive resistance classifier using beta-lactamase gene sequences and other genomic markers that are linked to known treatment outcomes and strain phenotypes. We will develop this new classifier based on published methods found to be effective with HIV genotype-phenotype prediction. ESBL-resistant features obtained from the clinical samples in this study will be used for training and testing of the classifier. The powerful combination of sharable database and analytic tools in a single platform will significantly advance knowledge of antibiotic-resistant bacteria, facilitate epidemiological monitoring of the spread of ESBL resistance, and represents a key first step to develop a diagnostic tool to counter ESBL resistance using whole-genome sequences.!